Method and Apparatus for Augmenting Data in Monitoring Video

1. A method for augmenting image data in a monitoring video, comprising: selecting randomly, based on a function to be implemented for the image data and a parameter range which are set one of by a user and by default, one or more processing functions to be performed on the image data and parameters of the one or more processing functions; processing the image data using randomly selected one or more processing functions and randomly selected parameters, to obtain new image data with randomness and comprehensiveness as augmented image data; and repeating the selecting and the processing until a required amount of augmented image data is generated, wherein the one or more processing functions comprise: spatial transformation, illumination change, noise filtering, scene synthesis and event modification, wherein the selected one or more processing functions are performed in a random order, and when the selected one or more processing functions comprise the scene synthesis, the scene synthesis is first performed on the image data, and then processing functions other than the scene synthesis are performed in a random order.

2. The method according to claim 1 , wherein the spatial transformation comprises one or more of: image spatial stretching, image rotation, image cropping, and image distortion simulation.

3. The method according to claim 2 , wherein the image distortion simulation is to simulate images captured by different imaging apparatuses, and comprises: correcting distortion of an image based on a known parameter of an imaging apparatus to produce a corrected image; and applying a new imaging parameter to the corrected image.

4. The method according to claim 1 , wherein the illumination change comprises a random illumination change and a video scene-based illumination change, wherein the random illumination change comprises: in a color space comprising hue, randomly changing values of one or more color components other than the hue while maintaining the hue unchanged; and the video scene-based illumination change comprises: processing each frame of a video by: for an n-th frame of the video, selecting randomly a frame which one of precedes and follows the n-th frame and is spaced from the n-th frame by m frames as a reference; and applying histogram matching to one of an (n+m)-th frame and an (n−m)-th frame to obtain a new n-th frame, with n and m being any positive integers.

5. The method according to claim 1 , wherein the noise filtering comprises noise addition and out-of-focus simulation, wherein the noise addition comprises changing an original value of a pixel at a random position in an image with a random number; and the out-of-focus simulation comprises one of resetting values of one of all and a part of pixels in the image calculated using a filtering function based on values of neighboring pixels and performing out-of-focus simulation based on a camera parameter and an estimated object distance.

6. The method according to claim 1 , wherein the scene synthesis comprises: providing, through a material library, materials of an object and a scene which exist in the monitoring video; simulating various actions; and organizing the materials from the material library to generate a virtual monitoring image.

7. The method according to claim 1 , wherein the event modification comprises: event deletion and event addition, wherein the event deletion comprises: detecting an event, deleting an object related to the event, and repairing a scene in which the object is deleted to generate a new image; and the event addition comprises: extracting scene information, specifying an event range, and adding an event using a function from the scene synthesis.

8. An apparatus for augmenting image data in a monitoring video, comprising: a function selector configured to select randomly, based on a function to be implemented for the image data and a parameter range which are set one of by a user and by default, one or more function sub-units for processing the image data and parameters for the one or more function sub-units; a processor, configured to process the image data using randomly selected one or more function sub-units and randomly selected parameters, to obtain new image data with randomness and comprehensiveness as augmented image data; and a repetition controller, configured to repeat processing performed by the function selector and the processor, until a required amount of augmented image data is generated, wherein the function sub-units comprise: a spatial transformation sub-unit, an illumination change sub-unit, a noise filtering sub-unit, a scene synthesis sub-unit and an event modification sub-unit, wherein processing of the selected one or more function sub-units is performed in a random order, and when the selected function sub-units comprise the scene synthesis sub-unit, the image data is first processed using the scene synthesis sub-unit, and then processing of the function sub-units other than the scene synthesis sub-unit is performed in a random order.

9. The apparatus according to claim 8 , wherein the spatial transformation sub-unit is configured to perform one or more of: image spatial stretching, image rotation, image cropping, and image distortion simulation.

10. The apparatus according to claim 9 , wherein the image distortion simulation is to simulate images captured by different imaging apparatuses, and comprises: correcting distortion of an image based on a known parameter of an imaging apparatus; and applying a new imaging parameter to a corrected image.

11. The apparatus according to claim 8 , wherein the illumination change sub-unit is configured to perform one or more of a random illumination change and a video scene-based illumination change, wherein the random illumination change comprises: in a color space comprising hue, randomly changing values of one or more color components other than the hue while maintaining the hue unchanged; and the video scene-based illumination change comprises: processing each frame of a video by: for an n-th frame of the video, selecting randomly a frame which one of precedes and follows the n-th frame and is spaced from the n-th frame by m frames as a reference; and applying histogram matching to one of an (n+m)-th frame and an (n−m)-th frame to obtain a new n-th frame, with n and m being any positive integers.

12. The apparatus according to claim 8 , wherein the noise filtering sub-unit is configured to perform one or more of noise addition and out-of-focus simulation, wherein the noise addition comprises changing an original value of a pixel at a random position in an image with a random number; and the out-of-focus simulation comprises one of resetting values of one of all and a part of pixels in an image calculated using a filtering function based on values of neighboring pixels and performing out-of-focus simulation based on a camera parameter and an estimated object distance.

13. The apparatus according to claim 8 , wherein the scene synthesis sub-unit is configured to: provide, through a material library, materials of an object and a scene which exist in the monitoring video; simulate various actions; and organize the materials from the material library to generate a virtual monitoring image.

14. The apparatus according to claim 8 , wherein the event modification sub-unit is configured to perform one or more of: event deletion and event addition, wherein the event deletion comprises: detecting an event, deleting an object related to the event, and repairing a scene in which the object is deleted to generate a new image; and the event addition comprises: extracting scene information, specifying an event range, and adding an event using a function from the scene synthesis.

15. A non-transitory computer readable storage medium storing a program which, when being executed by a computer, causes the computer to implement the method according to claim 1 .